The development of radiation predictive assays has been a central question in radiation research for more than fifty years. Recent developments in genomic profiling (DNA microarrays) have led to promises that predictors or classifiers will be developed to predict the classification and prognosis of cancer. We have developed for the first time a radiation classifier that predicts radiation sensitivity based on gene expression profiles derived from DNA microarrays. One of our goals is to improve the accuracy of our classifier which is currently 60%. This would allow us to eventually validate this new approach in a clinical trial. To do this we will expand the database of cell lines in the classifier to include all 60 cancer cell lines from the NCI. The design of our classifier involves two steps: a gene filtering step, where predictive genes are identified, and an evaluation process, where the predictor is tested. The gene filtering step is performed using significance of microarrays analysis (SAM), and the evaluation is performed using leave-one-out cross-validation. Our novel approach to gene identification consistently selected three known and one unknown genes as predictive of radiation response. In this proposal we plan to validate the results of this in silico analysis, by quantitative RT-PCR. Validation of these genes would not only confirm the validity of our analysis, but would establish three new genes that may be involved in radiation response. Finally, we have hypothesized that RbAp48, one of the genes identified in our analysis, plays a direct role in determining radiation phenotype. We will overexpress (transfection) and/or downregulate (antisense oligos),RbAp48 in cell lines and determine whether the radiation response is altered. In summary, our proposal may significantly advance radiation oncology research in several areas. Improvement in the accuracy of the classifier will potentially allow for the first time the identification of patients with radiocurable tumors, prior to the delivery of therapy. Validation of the predictive genes, will possibly identify several novel molecular pathways in radiation response.